JP2006525716A - Digital filter structure - Google Patents

Abstract

A digital filter configuration for the filtration of a digital video signal, wherein the functions of a zoom filter, in the form of a low-pass filter, which is a polyphase filter, and of at least one peaking filter, in the form of a high-pass filter, are realized, wherein the functions of the two filters are realized in a combined filter (1-17) in an integrated circuit in a manner such that, for each phase of the zoom filter to be set, combined filter coefficients are calculated from filter coefficients stored for this phase of the zoom filter and from filter coefficients stored for the peaking filter, which combined filter coefficients are applied to the video data to be filtered in a filtration process, so that both filtration functions are executed with the combined filter coefficients in this filtration process.

Description

  The present invention relates to a digital filter structure for filtering digital video signals. Thereby, one zoom filter in the form of a low-pass filter that is a polyphase filter and at least one peaking filter in the form of a high-pass filter are provided.

  This type of structure is known from the prior art. Here, in these structures, generally, the zoom filter and the peaking filter are individually configured as separate filters. Basically, the peaking filter exhibits a filter characteristic to which a specific number of filter coefficients for setting a filter scale are given. The zoom filter is in the form of a polyphase filter. That is, the zoom filter can realize different filter characteristics according to different stored filter coefficients. This filter is in the form of a low-pass filter. Both filters generally have the function of processing the video signal so that the display of the video signal is zoomed in on the display, i.e. an enlarged display is obtained. Thereby, the zoom filter has a function of suppressing undesirable high frequency components. Zooming in involves a loss of definition that must be offset using a peaking filter.

A structure and method for filtering a video signal, in which both a zooming function and a peaking function are realized, is known from US Pat. No. 5,422,827. However, even with this known method, the possible filter characteristics stored are all individually stored, and for every newly set phase, from the stored value, the zoom filter There is a drawback that the filter coefficients for complex filtering are read from the filter coefficients stored for each phase from the filter coefficients of the peaking filter. Thus, the memory requirements for all possible filter coefficients are quite high.
US Pat. No. 5,422,827

  The object of the invention is to specify a structure of the kind described above in which the filtering function of the zoom filter and peaking filter is realized in the simplest way and there is a minimum memory requirement in each of the stored filter coefficients of the two filters. That is.

  This object is achieved according to the invention by the features described in claim 1.

  That is, in the digital filter structure for filtering the digital video signal, there is a function of one zoom filter that forms a low-pass filter that is a polyphase filter and a function of at least one peaking filter that forms a high-pass filter. And the function of the two filters is realized in the integrated circuit by a composite filter, so that for each phase of the set zoom filter, from the filter coefficients stored for this phase of the zoom filter, and From the filter coefficients stored for the peaking filter, a composite filter coefficient is calculated, and the composite filter coefficient is applied to the video signal to be filtered in the filtering process, so that in this filtering process the composite filter coefficient is Use A digital filter structure, characterized in that both the filter function is performed.

  In the digital filter structure according to the present invention for filtering a digital video signal, a zoom filter and a peaking filter in the form of a polyphase filter are realized by a composite filtering function. That is, a filter whose filter coefficient is a composite filter coefficient calculated from the filter coefficient of the zoom filter and the peaking filter is used.

  Since the zoom filter is a polyphase filter that can have 16 different settings, in some solutions of the prior art, for example, the total of these 16 phases together with the 3 filter coefficients of the peaking filter. Composite filter coefficients must be stored for each. This creates significant memory requirements. In contrast, in the solution according to the invention, the composite filter coefficients are calculated in real time. This takes into account which phase is set by the zoom filter. Thereafter, in the digital filter structure, the composite filter coefficient is calculated in real time from the filter coefficient of the setting phase of the zoom filter and the filter coefficient of the peaking filter and used in the composite filter. This eliminates the need to store composite filter coefficients for all phases of the zoom filter. It is only necessary to store the filter coefficients of the polyphase filter itself. Since the composite filter coefficients are larger in number than the filter coefficients of only the zoom filter, the storage space can be saved.

  Further, the filtering function is improved by the combination of the zoom filter and the peaking filter. Assuming that the filter coefficients are the same, when using the composite filter according to the present invention, a better peaking function can be obtained than using individually provided filters having exactly the same filter coefficients.

  According to one embodiment of the invention as claimed in claim 2, a further reduction of the storage space for the filter coefficients of the polyphase filter is achieved. The filter coefficient may have a negative value if applicable. In the case of prior art solutions, these negative values are specified in memory by specific sign bits. In the solution as claimed in claim 2, all stored filter coefficients are subject to an offset, so that negative values no longer occur. In this way, the filter coefficient that receives this offset is stored. This eliminates the need for memory bits for codes. The offset is canceled after reading the filter coefficients and before its application in filtering.

In a further embodiment of the invention as defined in claim 3, only the half-phase filter coefficients of the polyphase filter are stored for the polyphase filter. As a result, the memory requirement is further halved. This is possible because the various phases of the polyphase filter have the property that the filter coefficients are mirror-symmetric with each other. The filter coefficients of one phase of the polyphase filter give a substantial weight to the various values used for filtering. Different phases of the filter give different weights to these numerical ranges. In practice, these weightings advance the mirror symmetry in the various phases of the polyphase filter. Thus, as provided under claim 3, it is possible to obtain the filter coefficients of the unstored phases from the stored phases, so that when the total number of phases is q, P _{1 to} P Only _{q / 2} phases are stored. For example, if a set of coefficients for phase P _q-r has to be obtained, the filter coefficients for phase P _r may be used and their order reversed. This allows half of the filter coefficients to be calculated from the other half, halving the memory requirement for the filter coefficients.

  In a further embodiment of the invention as defined in claim 4, the composite filter coefficients can be advantageously calculated according to the formula specified in the claim. This produces a composite filter coefficient that is one less than the sum of the filter coefficients of the two individual filters. Here, the fact that the composite filter coefficients are calculated in real time and therefore does not need to be stored is important.

  The present invention will be further described with reference to the embodiments shown in the drawings, but the present invention is not limited to these embodiments.

  The filter structure shown in FIG. 1 in the form of a schematic block circuit diagram has the function of filtering digital video signals. Thereby, both a zoom filtering function including a low-pass filter and a peaking function including a high-pass filter are realized. In the structure according to the present invention, these two filtering functions are realized in a composite filter that operates using a combined filter coefficient (composite filter coefficient). The composite filter coefficient is calculated in real time according to the setting of the zoom filter which is a polyphase filter.

  FIG. 1 shows a memory 1 in which three filter coefficients are stored. These filter coefficients are the filter coefficients of the peaking filter. Further, an adjustable mixer 2 is provided, and the strength of the peaking function can be set by this mixer according to the external signal Peak. Thereby, the filter coefficient from the memory 1 can be changed once again to adjust the strength of the peaking function. The filter coefficients for the peaking function that have been changed again in this way are sent to the unit 3 for the calculation of the composite filter coefficients.

  A ROM memory 4 is also provided for storing filter coefficients in one phase of the zoom filter in the form of a polyphase filter.

  A RAM memory 5 is also provided in which specific filter coefficients assigned for a plurality of phases of the zoom filter in the form of a polyphase filter are stored.

  Thereby, the ROM 4 has a function of making a set of filter coefficients available first during the cold start of the digital filter. As the structure enters operation, additional sets of filter coefficients are written into the RAM memory 5 and are available for various phases. Data is read from the RAM memory 5 in the unit 6 in response to the external signal Ph. Unit 6 is for selecting the set phase according to these specific filter coefficients that are assigned and transmitted to unit 7 for the reconstruction of filter coefficients. If applicable, the filter coefficients are not completely stored in the memories 4 and 5, as will be described later. Therefore, the unit 7 can reconstruct the actual filter coefficients in accordance with the data arrangement caused by storing the filter coefficients. For this purpose, a signal MM is transmitted to the unit 7 indicating whether only half of the filter coefficients are stored in the memories 4 and 5. Further, a signal VO indicating whether or not the filter coefficient is stored together with the offset is transmitted to the unit 7. In addition, a signal SP indicating the total amount of phase applied is transmitted to the unit 7.

  The unit 7 calculates actual filter coefficients from the filter coefficients transmitted from the unit 6 according to the signals MM, SP, and VO. Initially, these are still just filter coefficients for the zoom filtering function. Unit 7 supplies these filter coefficients in a particular set phase to unit 3 for calculating composite filter coefficients.

  In the embodiment shown in FIG. 1, it is assumed that the zoom filter shows four filter coefficients for each phase and the peaking filter shows three filter coefficients for each phase. In this embodiment, the unit 3 for calculating the filter coefficients calculates six composite filter coefficients, the amount being less than the sum of the individual filter coefficients of the two filters. These filter coefficients are transmitted to six multipliers 11 to 16 provided in the filter structure. To each of these multipliers, one scan value S1 to S6 of the video signal to be filtered is transmitted. For example, when zooming and peaking of the video signal must be performed in the vertical direction, the scan values S1 to S6 are values of 6 pixels arranged vertically in the adjacent scan lines.

  In the multipliers 11 to 16, these pixel values have specific filter coefficients added to them. The values thus obtained are summed in the addition stage 17. In the downstream rounding stage 18, the obtained value is rounded off, and then this value is divided by a predetermined value in the downstream stage 19. As a result, the value is scaled to a desired numerical range. Since overshoot may occur as a result of the filtering function in a digital signal, a clipping stage 20 is also provided to eliminate these overshoots when the overshoot exceeds a predetermined value. Clipping stage 20 provides an output signal to the digital filter structure.

  One important advantage of the digital filter structure according to the invention is inherent in the fact that the calculation of the composite filter coefficients in the unit 3 takes place in real time depending on the phase of the set zoom filtering function. Thus, the composite filter coefficients need not be stored for all phases. All filter coefficients that actually need to be stored are non-complex, i.e. the individual filter coefficients of the two filtering functions. This also saves further storage space as will be described later.

ここで、Ｆ_１は、ズームフィルタの相の係数であり、ｘ_２はその量である。それに対応して、Ｆ_２は、一つの相だけを示すピーキングフィルタの係数であり、ｙはその量である。ｎは、ｎ＝ｘ＋ｙ−１から得られる複合フィルタの係数の量であり、ｋは、１からｎまで続く計算される係数の連続番号である。従って、各設定相毎に、この式から、割り当てられて組み合わせられたフィルタ係数の組を計算することができる。これについて、二つの実施例を参照して説明する。 The unit 3 for calculating the composite filter coefficients may advantageously perform its calculation according to the following formula:

Here, F ₁ is a phase coefficient of the zoom filter, and x ₂ is the amount thereof. Correspondingly, F ₂ is a peaking filter coefficient indicating only one phase and y is its quantity. n is the quantity of coefficients of the composite filter obtained from n = x + y−1, and k is the sequence number of the calculated coefficients that continue from 1 to n. Therefore, for each set phase, a set of assigned and combined filter coefficients can be calculated from this equation. This will be described with reference to two embodiments.

F ₁ = A, B, C, D where A is the first coefficient, B is the second coefficient, and so on.
F ₂ = E, F, G where E is the first coefficient, F is the second coefficient, and so on.

Factor 1: (k = 1, x = 4, y = 3)
z = 0 z = 1 z = 2
yz = 3-> G yz = 2-> F yz = 1-> E
k−y + 1 + z = −1−> 0 k−y + 1 + z = 0−> 0 k−y + 1 + z = 1−> A

→ G ^* 0 + F ^* 0 + E ^* A = E ^* A

Coefficient 1 = E ^* A

Coefficient 2: (k = 2, x = 4, y = 3)
z = 0 z = 1 z = 2
yz = 3-> G yz = 2-> F yz = 1-> E
k-y + 1 + z = 0-> 0 k-y + 1 + z = 1-> A k-y + 1 + z = 2-> B

→ G ^* 0 + F ^* A + E ^* B = F ^* A + E ^* B

Coefficient 2 = F ^* A + E ^* B

Using the corresponding application of the above equation, the following is obtained for the remaining coefficients:

Coefficient 3 = C ^* E + F ^* B + A ^* G
Coefficient 4 = D ^* E + C ^* F + B ^* G
Coefficient 5 = D ^* F + C ^* G
Coefficient 6 = G ^* D

F ₁ = E, F, G where E is the first coefficient, F is the second coefficient, and so on.
F ₂ = A, B, C, D where A is the first coefficient, B is the second coefficient, and so on.

Factor 1: (k = 1, x = 3, y = 4)
z = 0 z = 1 z = 2
yz = 4-> D yz = 3-> C yz = 2-> B
k-y + 1 + z = -2-> 0 k-y + 1 + z = -1-> 0 k-y + 1 + z = 0-> 0

z = 3
yz = 1-> A
k−y + 1 + z = 1−> E

D ^* 0 + C ^* 0 + B ^* 0 + A ^* E = E ^* A

Coefficient 1 = E ^* A

Coefficient 2: (k = 2, x = 3, y = 4)
z = 0 z = 1 z = 2
yz = 4-> D yz = 3-> C yz = 2-> B
k−y + 1 + z = −1−> 0 k−y + 1 + z = 0−> 0 k−y + 1 + z = 1−> E

z = 3
yz = 1-> A
k−y + 1 + z = 2−> F

D ^* 0 + C ^* A + B ^* E + A ^* F = F ^* A + E ^* B

Coefficient 2 = F ^* A + E ^* B

Using the corresponding application of the above equation, the following is obtained for the remaining coefficients:

Coefficient 3 = C ^* E + F ^* B + A ^* G
Coefficient 4 = D ^* E + C ^* F + B ^* G
Coefficient 5 = D ^* F + C ^* G
Coefficient 6 = G ^* D

  In these two embodiments, the filter coefficients are “shifted up and down” by a kind of superposition, which results in the composite filter coefficients 1-6 described above. These composite filter coefficients realize composite filtering, and both a zoom filtering function and a peaking filtering function are realized in the meantime.

  As a result of this calculation of the composite filter coefficients in the unit 3 of the structure according to FIG. 1, storage space can be saved considerably. This is because these composite filter coefficients do not need to be calculated and stored in real time depending on the phase setting.

  As described above, filter coefficients for the various phases of the zoom filtering function are stored in the memories 4 and 5. FIG. 2 shows such a set of possible filter coefficients in 8 phases. In the example shown in FIG. 2, for the sake of simplicity, four filter coefficients are assumed for each phase. Therefore, as an example, in the example shown in FIG. 2, the filter coefficients in phase 0 are −6, 34, 215, and 13.

  Basically, there is an option to store the filter coefficients in the memory 5 in a pattern as shown in FIG.

  However, in the filter structure according to the present invention, the storage space can be further reduced. For this purpose, the set of filters shown in the arrangement of FIG. 3 is in principle the same as the arrangement in FIG. 2, but has undergone an offset, so that no filter coefficients are less than zero. As the filter data set of FIG. 2 shows, the smallest negative value that occurs is a value of -6. If the filter coefficient of the data set shown in FIG. 2 receives an offset of +6, a set of filter coefficients as shown in FIG. 3 is obtained. In the filter data set shown in FIG. 3, negative filter coefficients no longer occur. If these filter coefficients are stored in the version shown in FIG. 3, there is no longer a need to provide a sign bit for storage in this way. Thereby, the storage space for each filter coefficient is saved.

  As shown in the arrangement of FIG. 3, the filter coefficients are arranged in mirror symmetry with each other. In this case, there is no reversal of the order of the filter coefficients. For example, comparing the phase 0 filter coefficients with the phase 7 filter coefficients, if this inversion is applied, these filter coefficients are identical. The same applies to phases 1, 6, phase 2, 5, and phases 3, 4. Therefore, it is sufficient to store the filter coefficients of phases 0 to 3, as the schematic diagram of FIG. 4 shows. At this time, the filter coefficients in the phases 4 to 7 can be calculated from the phases 0 to 3. In this case, the principle described above is applied in the calculation. As a result, in the memory 5, it is possible to realize further reduction, that is, halving of the required memory requirement.

  Overall, as a result of complex filter coefficient calculations performed in real time depending on the phase of the zoom filter settings, and as applicable, as a result of further reduction of data in various phases of the zoom filtering function. A significant reduction in memory requirements is achieved with the digital filter structure according to the invention. This is particularly important as a practical problem. This is because the storage space requires a relatively large chip surface in the integrated circuit.

1 shows a schematic diagram of a digital filter structure according to the invention with a composite filter. An example of a set of filter coefficients in a polyphase filter is shown. An example of a set of filter coefficients in a polyphase filter is shown. An example of a set of filter coefficients in a polyphase filter is shown.

Claims (4)

  In the digital filter structure for filtering digital video signals, the function of one zoom filter in the form of a low-pass filter that is a polyphase filter and the function of at least one peaking filter in the form of a high-pass filter are realized. The function of the two filters is realized in the integrated circuit by a composite filter, so that for each phase of the zoom filter to be set, from the filter coefficients stored for this phase of the zoom filter and the peaking filter From the stored filter coefficients, a composite filter coefficient is calculated and applied to the video signal to be filtered in the filtering process, thereby using the composite filter coefficient in the filtering process. Both Digital filter structure, characterized in that the filtering function is performed.   The range of coefficient values is subject to an offset and negative values no longer occur, so the sign bit is not stored and the offset is canceled after the filter coefficient is read and before its application in the filter. 2. The digital filter structure according to claim 1, wherein filter coefficients in the phase filter are stored. If r is the phase sequence number and q is the total number of phases, the filter coefficients are stored for (q / 2) of the desired q phases and the filter coefficients in the remaining phases are stored. as calculated from the filter coefficients, data reduction of the x-number of filter coefficients in the polyphase filter is applied, thereby, in order to calculate a set of coefficients of a phase P _q-r, the stored phase p _r The digital filter structure of claim 1, wherein the filter coefficients are used to reverse their order. The calculation of the filter coefficient of the composite filter is as follows:

Done according to
F ₁ = Zoom filter phase coefficient x = Zoom filter phase coefficient F ₁ quantity F ₂ = Peaking filter coefficient y = Peaking filter coefficient F ₂ quantity n = Composite filter coefficient k = Composite filter The sequence number of the coefficients to be calculated (1, 2, ... n)
And
n = x + y−1.
2. The digital filter structure according to claim 1, wherein:

Images